Creating the Right Environment: Appropriate Use of Antibiotics in an Era of Resistance

The Economic Impact of Lower Respiratory Tract Infections

Sandra K. Willsie, DO, FCCP (University of Health Sciences, Kansas City, MO) provided extensive epidemiologic and economic statistics regarding infectious respiratory diseases. According to a World Health Organization (WHO) report in 1999, acute respiratory infections resulted in as many deaths as did AIDS and malaria combined. Acute respiratory infection may occur at any age and is most frequent in individuals 70 years of age or older; but from age 50 years on, the frequency rises steadily through the years in which comorbidities become more prevalent. 

In the United States, the highest economic cost of this group of diseases is associated with acute exacerbations of chronic bronchitis (AECB) related to chronic obstructive pulmonary disease (COPD). In one study, the mean cost per episode, duration of hospitalization, and time between episodes all depended upon treatment response to various regimens. Hospitalization accounted for the vast majority of total costs (Niederman MS et al. Clin Ther. 1999;21:576).

Community-acquired pneumonia (CAP) strikes between 2 and 3 million Americans annually resulting in approximately 10 million physician visits. Two hundred fifty-eight individuals per 100,000 population require hospitalization for CAP annually (962 per 100,000 in the population 65 years of age and older). Moreover, CAP is the nation’s sixth leading cause of death, claiming 14% of hospitalized patients (Bartlett J et al. Clin Infect Dis. 2000;31:347). 

From these and other data, Dr. Willsie concluded that “lower respiratory infections can result in significant need for hospitalization, expenditure of large sums of healthcare dollars, loss of productivity, significant mortality and, importantly, long periods of recuperation before patients return to normal.” 

The Etiology of Community-Acquired Respiratory Tract Infections

The etiologic pathogens associated with lower respiratory tract infections (LRTI) have changed in prevalence over time. While S. pneumoniae remains the most common causative pathogen of community-acquired pneumonia (CAP), a number of newer pathogens, such as C. pneumoniae and Sin nombre virus (hantavirus), have been recognized in recent years. Other commonly identified pathogens include H. influenzae, M. pneumoniae, C. pneumoniae, Legionella species, and viruses. S. aureus and Gram-negative bacilli are the cause in selected patients. The frequency of other etiologies, e.g., TB, Chlamydia psittaci (psittacosis), Coxiella brunetii (Q fever), Francisella tularensis (tularemia) and endemic fungi (histoplasmosis, coccidioidomycosis, blastomycosis) varies with epidemiological setting. In virtually all patient series reported in the 1990s, S. pneumoniae ranked first in frequency. In 30 to 60% of cases of CAP, however, no specific etiologic agent is detected. The incidence of mixed infections is unclear, although incidence rates ranging from 2.7% to 39% have been reported. In studies limited to cases with defined etiology, however, the frequency rate of mixed infections appears to be approximately 2 to 10%. Mixed infection may be either concurrent or sequential. It is probable that most are sequential, with the isolation of a bacterium such as S. pneumoniae occurring after serologic evidence of a respiratory virus, Mycoplasma, or C. pneumoniae. 

Thomas M. File Jr., MD, FCCP (Northeastern Ohio Universities College of Medicine) and his colleagues have observed, however, that with the exception of S. pneumoniae, the frequency patterns of pathogens vary with severity of disease as evidenced by site of care (community, hospital, or intensive care unit). Among ambulatory patients, S. pneumoniae is followed by M. pneumoniae, H. Influenzae, C. pneumoniae, and viruses. Among hospitalized patients not in the ICU, M. pneumoniae, C. pneumoniae, H. influenzae, Legionella species, and aspiration follow S. pneumoniae. In the ICU, S. pneumoniae is followed by H. influenzae, Legionella species, Gram-negative bacilli, and S. aureus (File TM et al. Curr Opin Pulm Med. 1997;3:89).

Genetic analysis using DNA amplification or polymerase chain reaction (PCR) is a new technique that should be helpful in the future to determine the etiology of LRTI. Older and more familiar methods can be used to characterize the etiologic diagnosis as definite, probable, or possible. The diagnosis is considered definitive if there is a positive blood or pleural fluid culture; a four-fold increase in antibody to a designated pathogen; isolation of H. influenzae, Legionella species, or tuberculosis from respiratory secretions; a positive urinary antigen for Legionella; or isolation of pneumococcus. Isolation of a Gram-positive bacterium from respiratory secretions constitutes a probable diagnosis. A diagnosis of CAP is possible if a pathogen is isolated from respiratory secretions or there is a high single antibody titer to a specified pathogen (Marston BJ et al. Arch Intern Med. 1997;157:1709). 

There is considerable variation in mortality rates associated with different respiratory pathogens in CAP. In a meta-analysis of more than 100 studies totaling approximately 20,000 patients, 13% of 11,229 patients in whom etiology was not defined died. This compared with 12% of 4,432 patients with S. pneumoniae, 7% of 833 patients with H. influenzae, 15% of 272 patients with Legionella species, 31% of 150 patients with Staphylococcus species, and 1.4% of 507 patients with M. pneumoniae (Fine M et al. JAMA. 1996;275:134). In this study, 4% of patients had evidence of mixed bacterial infections, and the mortality rate among them was 24%. 

Among outpatients, pneumococcus, Haemophilus, and Mycoplasma are the major causes of CAP based on a combination of definite and presumptive identification. In a study of outpatient CAP that correlated etiologic pathogens with severity using the Fine Index, however, atypical pathogens led by M. pneumoniae were the dominant organisms among younger patients with no significant comorbidities (Falguera M et al. Arch Intern Med. 2001;161:1866). Although C. pneumoniae has been observed in CAP with increasing frequency over the last decade, its role remains unclear. 

Bacteria are generally considered to cause approximately half of COPD-related acute exacerbations of chronic bronchitis (AECBs). (The fact that bacterial and nonbacterial cases do not manifest differentially poses a challenge with respect to empiric antibiotic therapy.) In recent studies, bacterial pathogens were isolated from sputum in 40 to 50% of exacerbations. When bacteria were involved, the core pathogens were H. influenzae (30%), S. pneumoniae (14%), and M. catarrhalis (14%). Complicated cases typically involved enterobacteriaceae, and Gram-negative organisms led by Pseudomonas species are common in severe exacerbations. Chlamydia species may be involved in 5 to 10% of exacerbations, and viruses in between 33% and 52% of cases (Sethi S. Clin Pulmon Med. 1999;6:327; Sethi S and Murphy TF. Clin Microb Rev. 2001;14:336). 

When, What, and Whether to Prescribe Antibiotics in LRTI

Daniel M. Musher, MD (Baylor College of Medicine) presented two cases that illustrate problems associated with empiric therapy in acute LRTIs. 

The first case was that of a generally healthy 53-year-old male who presented with symptoms and signs of CAP. In two previous episodes, H. influenzae had been identified. Based on his history, physical, and condition, he was treated with doxycycline as per the treatment algorithm of the ATS. Thirty-six hours later he was acutely ill and was hospitalized and switched to ceftriaxone and a quinolone. Sputum cultures subsequently revealed infection with doxycycline-resistant S. pneumoniae. Dr. Musher said that the important conclusion is that empiric therapy is inherently a gamble on the odds, and that the best diagnosis is an established diagnosis. He emphasized that the main lesson is NOT to use broader-spectrum antibiotics in empiric therapy; a Gram stain on this patient’s sputum would have revealed pneumococci and treatment with amoxicillin would have been given, presumably with a good response. 

Dr. Musher’s second case was that of an acutely ill middle-aged man who was transported to an urgent care center with CAP confirmed by chest X-ray. His infection was refractory to clindamycin plus gatifloxacin given empirically. His condition deteriorated, necessitating transfer to the MICU where he received vancomycin plus piperacillin/ tazobactam, again empirically. When his sputum was finally submitted, a Gram stain suggested refractile bodies that were shown to be acid-fast, leading to the diagnosis of tuberculous pneumonia. Dr. Musher concluded that early acid-fast staining is a casualty of empiric antibiotic therapy; and had it been done at the time of admission, the patient might have been treated promptly with appropriate antimicrobial drugs, probably avoiding transfer to the MICU and intubation with its associated complications. 
 
  

Classification and Differential Diagnosis of Acute Exacerbations of Chronic Bronchitis

Sanjay Sethi, MBBS (State University of New York at Buffalo) began his presentation by identifying medical-legal issues, patient expectations, unreliable diagnostic tests, lengthy turnaround time, and the lack of sound clinical data as the principal obstacles to consistent accuracy in the differential diagnosis of LRTIs (acute bronchitis, AECB, and CAP). Accurate diagnosis is essential for balancing efficacious empiric therapy with avoidance of unnecessary antibiotic use. 

Recently published guidelines for differentiating acute bronchitis from AECB and CAP exclude individuals who have chronic obstructive pulmonary disease (COPD), congestive heart failure (CHF), immunosuppression, and history of heavy smoking. For other patients who present with acute cough illnesses of less than 3 weeks’ duration, with or without sputum production, acute bronchitis is the probable diagnosis outside influenza season. If such a patient has abnormal vital signs or an abnormal pulmonary examination, a chest X-ray is needed to rule out CAP. If the X-ray is positive, treat as pneumonia. If it is negative, treat as acute bronchitis. If there is no history of underlying disease, the vital signs are normal, and the pulmonary exam unremarkable, treat as acute bronchitis. 

Differentiating AECB from CAP is complicated by the fact that acutely ill hospitalized patients with COPD have positive X-ray findings in only about 7% of cases. Thus X-ray is indicated for these patients only when they also have coronary artery disease, CHF, chest pain, dementia, lung consolidation, pulmonary edema, or elevated white cell counts.

Because AECBs are bacterial in origin in only about half of cases, it is important to determine whether or not there is a bacterial infection. This can be done by four methods: Gram staining, culturing, observing gross purulence in sputum, and assaying sputum for inflammatory markers for bacterial infection. Gram staining and sputum cultures are of limited value because they cannot differentiate between colonization and acute infection. Furthermore, many patients start with viral infections and then develop bacterial superinfections that could not be detected by sputum cultures in early diagnosis. Mucoid sputum correlates highly with negative cultures and purulent sputum correlated highly with positive cultures. Mucopurulent sputum provides no diagnostic clues (Stockley RA et al. Chest. 2000;117: 1638). In research involving 81 exacerbations in 45 patients, assaying sputum for free neutrophil elastase appeared to be an effective way of determining the presence of H. influenzae and M. catarrhalis (Sethi S et al. Chest. 2000;118: 1557). Currently, however, methods and combinations of methods of detecting bacteria in AECB are only approximately 65% accurate, which is insufficient for making a decision to prescribe antibiotics. 

There is, however, clinically useful information about stratifying patients to assess the probable benefit of antibiotic therapy in AECB. In a study of 362 exacerbations in 173 patients, exacerbation was rated by patient reports of symptoms and symptom increases. Within each of three strata, patients were randomized to standard antibiotics or placebo. Overall, at the end of 3 weeks, there was a significant difference in recovery rates favoring antibiotics. Almost all of the difference, however, was between the sickest patients and those with intermediate symptoms, suggesting that those with most symptoms have the highest probability of benefiting from antibiotic therapy. In addition, twice as many patients taking placebo became sicker as did patients treated with antibiotics, again with maximum observed benefit in the patients with most symptoms (Anthonisen NR et al. Ann Intern Med. 1987;106:196). When patients are stratified by this method, antibiotics should be withheld from patients with the fewest symptoms. 

Antibiotics can also be stratified based on in vitro sensitivity to specific organisms and in vivo microbiologic efficacy at varying minimum inhibitory concentrations (MICs). 

 

Drug-resistant Organisms in LRTI: Do They Matter in Patient Management of CAP?

Michael S. Niederman, MD, FCCP (State University of New York at Stony Brook) noted that most of the growth in pneumococcal resistance is in the intermediate range rather than the high range. However, penicillin resistance co-exists with resistance to other antibiotic classes. Consequently “penicillin-resistant S. pneumoniae” (PRSP) is more accurately “drug-resistant S. pneumoniae” (DRSP). By this standard, 40% of pneumococci in the United States are resistant to penicillin.

Pneumococcal macrolide resistance is an increasing problem. This organism can resist this class of antibiotics by both efflux and ribosomal mechanisms, with efflux inducing a lower level of resistance than ribosomal mutation. In the United States, most of the increase in macrolide-resistant pneumococci is in serotypes that use the efflux mechanism, so many remain susceptible to oral macrolide therapy. Those strains that use the ribosomal mechanism are clearly not susceptible to macrolides (Gay K et al. J Infect Dis. 2000;182: 1417). 

The antibiotic resistance of H. influenzae and M. catarrhalis involves inactivating beta-lactamases. In a study of 1,537 H. influenzae isolates, 37% were associated with beta-lactamase production. An additional 2.5% were beta-lactamase negative, but they were resistant to ampicillin and amoxicillin via modified penicillin-binding proteins (Doern GV et al. Antimicrob Agents Chemother. 1997; 41:292). With M. catarrhalis the problem is even more severe. In a study of 723 isolates, 95% produced beta-lactamase (Doern GV et al. Antimicrob Agents Chemother. 1996;40: 2884).

The newest ATS treatment algorithm for CAP takes antibiotic resistance into account. Specifically, it includes modifying risk factors for infection with drug-resistant organisms (Table 1). The ATS treatment recommendations for patients at risk for DRSP in CAP, excluding ICU patients, appear in Table 2. 

Macrolides do not appear to be a problem provided they are limited to populations without risk for DRSP, Gram-negative organisms, or aspiration. Quinolones provide coverage for Gram-positive, Gram-negative, and atypical organisms and they penetrate secretions well. However, a study of 7,551 pneumococcal isolates demonstrated that quinolone resistance increases with the amount of quinolone use and that penicillin resistance and quinolone resistance can co-exist (Chen DK et al. N Engl J Med. 1999;341:233). Therefore, if quinolones are used for the treatment of CAP, the most potent members of the class (gatifloxacin, moxifloxacin, and gemifloxacin) should be selected. 

The use of vancomycin should be limited to patients with meningitis, documented high-level resistance, and infection refractory to other antimicrobial agents. 

The Impact of Bacterial Resistance on Clinical Outcomes in Real Patients

To illuminate the relationship between in vitro resistance and clinical response, John R. Lonks, MD (Brown Medical School) presented data on treatment failures of patients with pneumococcal respiratory infections taking tetracyclines, macrolides, quinolones, or cephalo-sporins. Patients who failed treatments were receiving antibiotics to which their respective S. pneumoniae isolates were resistant. Many of the pneumococcal isolates were from sputum, but at least one patient receiving a tetracycline, macrolide, quinolone, or oral cephalosporin had a resistant pneumococcus isolated from blood. To the extent possible from the published reports, Dr. Lonks compared the MIC range of etiologic organisms with the peak concentrations of the agents used. In each case he found that treatment failure was associated with resistance levels that were substantially out of the peak concentrations of the agents selected for therapy.

On the basis of these findings, Dr. Lonks concluded that treatment failures of pneumococcal respiratory tract infections to four classes of antibiotics “highlight the clinical relevance of in vitro resistance.” 

All contents Copyright © 1999 - 2002 Medical Association Communications